Scaling fundamentals
Latency, throughput, load balancing, stateless services, CAP, and the first bottlenecks that appear under traffic.
RunDrill plays the interviewer, pushes on weak tradeoffs, scores the answer, and brings the same gaps back until they stop leaking.
Free while in beta · works inside Claude, Codex, Antigravity, and any MCP client.
Review plausible designs with planted defects: hot shards, lying caches, unbounded queues, wrong consistency, or one critical service with no backup.
The coach pushes back like an interviewer. You commit, explain what you gain, name what you lose, and adapt when the requirement changes.
Back-of-envelope estimates become short drills, not a vague interview ritual. You learn when the estimate changes the architecture.
Mock interviews end with a scorecard across problem navigation, solution design, technical excellence, and communication.
The course climbs five bands: foundations, components, patterns, complete system designs, and interview execution. Each miss becomes future practice instead of disappearing into chat history.
Latency, throughput, load balancing, stateless services, CAP, and the first bottlenecks that appear under traffic.
SQL and NoSQL, indexes, replication, sharding, caches, queues, streams, object storage, search, APIs, and WebSockets.
Design rate limiters, feeds, chat, video, ticketing, search autocomplete, payments, collaborative editing, cloud storage, and more.
Clarify requirements, estimate capacity, draw the high-level design, choose deep dives, discuss tradeoffs, and close cleanly.
A session is not a lecture. The coach asks one thing, waits for your answer, then probes the weak spot before revealing the stronger version.
It reads your current level, open mistakes, target track, and recent weak spots before picking the next drill.
You clarify, estimate, critique, design, or defend first. The coach does not hand you the architecture before you try.
If the answer was hard, the close includes a compact ideal answer so you can see the shape: scope, architecture, tradeoffs, bottleneck, and next deep dive.
Today’s gap: you named components but did not say which requirement each one protects.
Add the RunDrill marketplace, install the system-design plugin, then start the coach inside the agent where you already work.
Type these straight into Claude Code or Claude Desktop:
# add the marketplace, then the course /plugin marketplace add rundrill/rundrill /plugin install rundrill-system-design@rundrill # then start coaching /system-design-coach
Add the RunDrill catalog, then install from the plugin directory:
codex plugin marketplace add rundrill/rundrill
Install rundrill-system-design, then start the system-design-coach skill.
Drop the rundrill-system-design folder into one of these:
# for every project ~/.gemini/config/plugins/rundrill-system-design/ # just this workspace <workspace>/.agents/plugins/rundrill-system-design/
Restart the agent and start the system-design coach.
Any MCP client can point directly at the coach endpoint:
{
"mcpServers": {
"rundrill-system-design": {
"type": "http",
"url": "https://mcp.rundrill.com/coach/system-design"
}
}
}
First run opens a browser tab to sign in once. Your progress lives on the RunDrill server, so weak spots follow you across chats and machines.
System Design is one of the RunDrill programs — languages, programming, interview prep, and exam prep, all inside your agent.